CP Lab 5: Functions, pointers, some arrays

Transcription

1 Computer Programming (CP) Lab 5, 2017/18 1 CP Lab 5: Functions, pointers, some arrays Instructions The purpose of this Lab is to help you get experience in understanding parameter-passing functions in C, and also understanding how arrays and pointers are related. You already have experience of using functions from your work with the descartes library in Lab 4. In this lab we will explore how to design functions, as well as to use them. All Labs for CP will take place on the DICE machines in AT5.05 (Thursday and Tuesday) or AT6.06 (Monday). For working alone, you may use any of the DICE labs in AT, as long as they are not booked out by a class. You have 24-hour access to AT levels 3,4,5: if your card/pin does not allow you in, use the ITO contact form on our course webpage to contact the ITO. Aims We will experiment with the details of the rotate.c function from Lecture 9. Explore the role of pointers as parameters to functions. In particular we will do some experiments with different versions (correct, and incorrect) of swap functions. We will show you how to print out the memory address of a particular variable using printf with the %p indicator. We will extend the concept of program planning to take in the design of functions. We will learn how arrays work when being used as parameters to a function. We will learn the difference between arrays (static pointers) and standard (dynamic) pointers. Prerequisites You should have attended your scheduled Lab sessions in weeks 2 5, and completed the Lab sheets during that session. You should be up to date with the material from the CP lectures. NOTE: If you are still behind with lab 4 (Descartes), then please do Stage A only of this lab, then return to working on lab 4. Come back to stages B and C of this lab later. If you are fully caught up, then do this lab in order. Getting started Log into on of the DICE terminals, bring up a Terminal window, move to your home directory, and create a new subdirectory called lab5. You may want to copy the template.c file from your lab2 subdirectory into the lab5 directory. 1 1 If you don t know how to do this, look back at the Lab sheet for Lab 3.

2 Computer Programming (CP) Lab 5, 2017/18 2 Stage A Arrays as parameters (rotate) Towards the end of Lecture 9, we discussed how functions can take arrays as parameters, by adding a declaration like int a[] (or double reals[], etc) in the header of the function prototype. We discussed how we must include an int parameter for each array, to pass the length of the array (C passes an array into a function as a pointer, so the function does not automatically know the length of the array). On slide 15, we gave code for a function called Rotate, whose input parameters are an int array and also an int (interpreted as being the length of the array). The function performs a single clockwise rotation of the elements of the array. In this section of the lab we will experiment with rotate.c to understand why the result of the rotation can be seen in the original array in main after the function terminates. The short answer is that Rotate is called with an address (the starting address of the array), and therefore Rotate will then use this address to access and manipulate the cells of the array starting there (stored in main s environment). We will now create a program rotate.c to illuminate this explanation. Copy template.c to make a new file called rotate.c and then add the code for the Rotate function (slide 15 of lecture 9) as a function above main. Please also add a printf inside the Rotate function, to print the start address of the b array, and indicate this output is coming from inside Rotate. Use %p to output an address with printf. At the start of main, define a local int array of length 10 named primes, and initialise it to store the first 10 prime numbers (see lecture 9 for how to do this). Next, add a printf to main which prints the start address of the primes array, and indicates this output is coming from main. Again, use %p to output an address with printf. After this, add the following call in main: Rotate(primes, 10); After this, use printf (probably in a loop) to output the sequence of entries in primes. After this, add the following call in main: Rotate(primes, 5); and then again output the sequence of entries in primes. Now compile your code as follows: gcc -Wall rotate.c After DEBUGGING is finished, run your program. Your working with rotate.c You should be able to observe the following behaviour from running your program: The address printed from within Rotate should be the same as that printed from main. After the call Rotate(primes, 10), the array primes should store the following sequence: 29, 2, 3, 5, 7, 11, 13, 17, 19, 23 After the extra call Rotate(primes, 5), now primes should store the following sequence: 7, 29, 2, 3, 5, 11, 13, 17, 19, 23 Do you fully understand why the program gave these results? If not, ask the Demonstrator or Lecturer! Now submit your program with submit cp lab5 rotate.c

3 Computer Programming (CP) Lab 5, 2017/18 3 The remainder of this lab takes you through a series of experiments to illustrate pointers and arrays in more detail, and give you a fuller understanding. If you are not confident with the course material so far, you should skip the rest of this lab, continue catching up with the previous labs, and come back to this point later if you have time. Stage B Swap Can we write a function that takes two variables and exchanges their contents? This function does not work: void swapwrong(int a, int b) { int tmp; tmp = a; a = b; b = tmp; } int i = 2, j = 7; swapwrong(i,j); // does not swap i and j To change variables that don t belong to to a function, we have to use pointers, like this: void swap(int *aptr, int *bptr) { int tmp; tmp = *aptr; *aptr = *bptr; *bptr = tmp; } int i = 2, j = 7; swap(&i,&j); // does swap i and j Now in this section of the Lab, we ll explore why the first does not work, and why the second does. Step 1 In this part of the Lab, we are going to code up both of the swap functions in the same program for testing. Your program should be named swap.c you can start it off by making a copy of template.c. The swap.c program will contain the two swap functions together with a main which will declare two local variables (local to main) i and j with values 1 and 2 respectively; and which will then ask the user which swap function to use; and then output the values of i and j after calling that function. This will allow us to check which version of swap worked and which does not. In Step 2, we will add some extra printf statements to understand why one swap works and the other does not. We are going to define a program which will have the following features: We will declare and implement the non-pointer swapwrong. We will declare and implement the pointer swap. Next we must write the initial code for main

4 Computer Programming (CP) Lab 5, 2017/18 4 The first thing to appear in main will be the declaration of two local int variables i and j, initialised to 1 and 2 respectively. Next main will use a printf to ask the user if they want to use swapwrong or swap, followed by a scanf to read in their response (0 meaning swapwrong, 1 meaning swap, any other int value causing an error message). Notice that to do this, we will need an extra variable to read in the option chosen by the user - so we will need to go back and define another local int variable - maybe called option, say... Also remember we need to use &x when using scanf with the variable x. We will use an if-statement to either Call swapwrong with the values of i and j, or Call swap with the addresses of i and j Note that since swap takes addresses as its parameters, we must use &i and &j in the call. Finally, we will use a printf to output the new values of i and of j in that order. This will allow the user to test whether the swap succeeded or not. I have drawn up the program plan for swap.c, and it appears over on the next page. Please use this as the basis to code up this first version of swap.c After you have written the program, please compile it using gcc -Wall swap.c Then run it (by typing./a.out in your Terminal window), first asking it to use swapwrong, and on the next run, to use swap. Verify that no swap happens when swapwrong is used, but we do get a swap using swap. Step 2 Now in Step 2, we will understand why swap achieves the swap but swapwrong does not. The reason is due to the mechanism of call by value which is the paradigm always used by C in substituting for parameters to functions. For swapwrong the following is the situation: The declaration (or function prototype) of swapwrong asks for two input parameters of type int, locally named a and b Therefore when we are using swapwrong in main we must pass it two integers - specifically for us, these will be the values of i and j. The call by value mechanism copies the values of i and j and puts these into the boxes for local variables a and b, and then the body of swapwrong executes on these local variables. For swap the situation is different: The declaration (or function prototype) of swap declares two input parameters which are the memory addresses of two int variables. These input parameters are locally named a and b. If we want to examine the values stored at the address a, we will have to write *a (for example).

5 Computer Programming (CP) Lab 5, 2017/18 5 Figure 1: Program plan for swap.c for you to complete.

6 Computer Programming (CP) Lab 5, 2017/18 6 When we are using swap in main we must pass it two memory addresses where integers are stored - specifically for us, this will be &i and &j. The call by value mechanism is still used for swap, but the values to be copied are the values of the address &i and the address &j. These address values get copied into the boxes for the local variables a and b of swap. The body of swap will execute wrt these address values. Because swap has been passed the addresses of the original variables of interest from main, the swapping gets done on these original locations. We now show how to add printfs to main, to swapwrong and to swap to show us which memory addresses are being worked on. Variables of main We will need to know the memory locations of the original variables i and j in main, so that we can check whether swapwrong and swap work with respect to these locations or not. Therefore in main, just underneath where we have our variable declarations, please add the following line of code 2 : printf("i and j are stored at addresses %p and %p.\n\n", &i, &j); Variables of swapwrong To illuminate what is going on with swapwrong, we will get it to tell the user the memory locations of the values that it swaps. Add the following lines of code at the start of the body of swapwrong. printf("\nwe are in swapwrong.\n"); printf("we are swapping the values at addresses %p and %p.\n\n", &a, &b); The value of adding those lines is that before swapwrong does its work, it will tell the user which memory locations are being worked on. Variables of swap We also want to understand why swap does work. We know that it has been passed addresses of memory locations, not values. Therefore we will print out both the addresses and also the values stored there. Add the following code at the start of the body of swap. printf("\nwe are in swap.\n"); printf("we are swapping the values at addresses %p and %p.\n\n", a, b); Comparing the two swaps Now recompile your code after having added these outputs: gcc -Wall swap.c There may be some DEBUGGING to be done. 2 We use %p to indicate an address is to be passed for output. The substituted value must either be a pointer variable (eg, x, if x was declared as a pointer variable) or alternatively must be the address &x of a variable.

7 Computer Programming (CP) Lab 5, 2017/18 7 Next, do two runs of the code (as usual, do this by typing./a.out at the command line) In each run, you will see the information about i and js addresses; then the query about which swap to use. First ask for swapwrong to be used - that means that will be given the result of the three printfs you added to swapwrong. Please carefully examine the memory addresses, and compare them to those initially given for i and j. Do they match? Does this explain why swapwrong does not work? On the second execution, ask for swap - this time you will see the result of the three printfs you added to swap. Please carefully examine the memory addresses, and compare them to those initially given for i and j. Do they match? Does this explain why swap does work? Your result for swap.c From Step 1, you should see that no swap happens if swapwrong is used; but a swap does happen if swap is used. In Step 2, when working with swapwrong, you should notice that the addresses printed out by swapwrong are not the addresses of i and j. In Step 2, when working with swap, you should notice that the addresses printed out by swap are the same as the addresses of i and j. Are you happy you understand the reason that swap causes i and js values to be exchanged back in main. If not, ask the demonstrator or lecturer to explain. Can you now understand the reason scanf requires you to pass the address of the variable where the read-in value will be stored, even though printf just wants a variable name? When you are finished, rename a.out to swap or similar for later use. Submit your program with submit cp lab5 swap.c

8 Computer Programming (CP) Lab 5, 2017/18 8 Stage C Arrays and pointers This first task involves writing a short simple program to illustrate the difference between pointers and arrays. As mentioned in the Arrays lecture, it is true that arrays are pointers, however there are subtle differences. You will need to start a new program - called array.c or similar - you may want to use your template.c file to start off. Within your main, please start off by making just two declarations: int *p; int r[10]; What do these declarations achieve? (a) int *p; will declare one box (variable) in memory which has enough space to store a memory address (memory address of an int variable): (i) The initial value which sits in p will be just random junk. (ii) There is no guarantee that the 0xff... junk in p at the start, actually corresponds to any actual usable address in memory. (iii) The address stored in p can change: we can initialise it, reassign different addresses later, etc. (b) int r[10]; will declare 10 boxes in sequential order in memory, each with space to store one int, and will set the static pointer r to have the address of the first (r[0]) of these 10 boxes. (I) The address that r has just after this declaration does correspond to a true address (with space for one int) in memory. (II) r corresponds to a real specific address where 10 int variables start. It will be associated with this specific address for all its life (r is a STATIC pointer) One thing before we start investigating - after the declarations, please add the following assignment statement: r[0] = r[0]; This has no effect on values stored; we are only adding it to avoid getting a Warning about the array r being set but not used. 1: Demonstration of (a):(i)-(ii) and (b):(i) To demonstrate (a)(i) and (a)(ii), add the following lines to appear after the two declarations of array.c (the second line is only there so that gcc will not complain about r being an unused variable): *p = 6; r[6]=4; Now save this program, and return to the command line, and compile with gcc. You will probably see the following warning: array.c:7: warning: p is used uninitialized in this function Since the above is only a warning, maybe try running the code by typing./a.out. You will probably get a segmentation fault - A segmentation fault means trying to access memory which it is not allowed to, or which doesn t exist.

9 Computer Programming (CP) Lab 5, 2017/18 9 There is nothing wrong with the form (or syntax) of *p =6;. This is appropriate for any actual memory address representing a variable. Problem is that p is uninitialized and the address it has is just junk. Notice the complaint is only for p. There were no problems with r[6]=4;, because space for the variable r[6] (and indeed all of r[0],.., r[9] had been allocated with that declaration), and could be accessed in terms of r. TO FIX THIS PROBLEM, add the following declaration and assignment, just AFTER your initial two declarations and ABOVE the assignment statement *p = 6:... int a; p = &a; *p = 6; Now compile and run the program a second time. You will have no difficulties. This is because before we tried to do *p (to look at what is stored at address p) we gave it the address of a existing int variable, specifically, the address of a. So instead of having a junk address as its value, p now is pointing at a true memory address that we have access to. Experiment with printf: You can print out memory addresses with printf. In the special case of the zero address, which in C is always junk, and used to mean invalid pointer, printf will print it as (nil). To see this, add the following command to your code: printf("p points to the address %p\n", p); First add it immediately below the int *p; declaration; and the second time, to be after the assignment p=&a; 2: Demonstration of (a)(iii) and (b)(ii) Now we are going to test assignment of addresses between typical pointers, and arrays. Go to the end of your program, just above return EXIT_SUCCESS;. Add the following line of code:... r = p; /* Goal: give pointer r the address of pointer p */ return EXIT_SUCCESS; Compile with gcc, and this time you get an error, not just a warning: array.c:10: error: incompatible types when assigning to type int[10] from type int * The problem is that arrays are static pointers, and they are permanently tied to their initial storage allocation. You cannot change them to point to a different address in memory, not even to the address of another array (not even to the address of another array of the same length and type). Now delete the most recent command and change it to:... p=r; /* Goal: give pointer p the address of pointer r */ return EXIT_SUCCESS; Compile with gcc You should see everything is fine.

10 Computer Programming (CP) Lab 5, 2017/18 10 Add the following commands printf("r points to the address %p\n", r); printf("p points to the address %p\n", p); in two places: just before the assignment p=r;, and just after it. Compile again and examine the output. The lesson you can take away is that standard pointers (declared as int *a or float *f, or whatever type) are re-assignable with memory addresses of any variable of that type. Array pointers are static and cannot have their value changed - though the value of what they point to can be changed. 3: Array indexing and pointers We all know by now certain relationships between arrays and pointers: r, the address of the array, is the same as &r[0] r[3] (say) is equivalent to writing *(r+3). In this expression r+3 means 3 boxes up from address r. Then the * means go 3 boxes up from address r and return the (int) value there. r[3] is just shorthand for *(r+3). We know that for a standard (non-array) pointer p we can assign p=r; to give p the address of an array. Some more interesting things: If we want to, we can assign p to point to an address in the middle of an array, for example: p = (r+3); Or equivalently, p = &r[3]; After having done this, we can use p[0] and p[1]... p[6] to access successive locations after p[0] (which is r[3]). Because gcc does not do out-of-bounds checking for arrays, we can also do p[7],..., p[9] (these are not part of r, as p started 3 boxes up), and even p[10], p[11] etc. Try it! Print out the values at these addresses and see what junk you get (maybe initialise all values of r first to see the difference). In fact C even lets you use p[1] etc even if the pointer variable p was only set to a standard non-array address. Try this out by adding the following commands at the end of your program:... p = &a; printf("look two-boxes-up from p: address %p, value %d.\n", &p[2], p[2]); return EXIT_SUCCESS; Lessons to take away are that C is very flexible with array indexing, even allowing it to be used with pointers which have never been related to arrays and (unfortunately) C does not check array-bounds.

11 Computer Programming (CP) Lab 5, 2017/18 11 Your working with array.c From part 1, should understand that int *p; pointers come initialised to junk values, but that array pointers correspond to real accessible addresses in memory. From part 2, should understand that an array, despite being a pointer, is a static pointer tied to a particular location in memory. From part 3, should understand the interesting relationship between arrays, pointers and *, and all that can be done with them. Also should understand pointer arithmetic (p+3) Submit your program with submit cp lab5 array.c Julian Bradfield after Mary Cryan, October 2016